[SOLVED] number of "Scalar Field" in ResultOutputSolve

[Martina]

Hi,

I just switched from r6158 to r6235, since them I'm not allowed to have more than 10 Scalar Fields in the ResultOutputSolver. Before it worked fine, now simulations with sif-files which have been working before end with messages like

Code: Select all

ERROR:: Model Input: 
ERROR:: Model Input:  Unknown specifier: [b95]
ERROR:: Model Input:  In section: [solver 16]
ERROR:: Model Input:  For property name:[scalar field 11]

I tested that this does not depend on the variable, it's always the number 11 which creates the error which ever it is. In the example, if b95 is number 10, it works fine. I also tried to put the same variable as 10 and 11, if there is only 10 it works, if there are 11 Elmer complains. I spent some time now browsing through the revisions of ModelDescription.src and ResultOutputSolve, but couldn't find so far anything related to this limitation.

Any ideas on that?

Thanks,
Martina

[gagliar]

Same issue with my install on a mac (version 6221).

Regards
Gag

[gagliar]

Hi,

In fact, my problem was not in the ResultOutputSolver but in the TwoMeshes solver when the keyword "Variable 11" was reached. I solved it by adding the type of the variable.

I guess that adding the type of the variable will also solve your problem? In your case:

Code: Select all

scalar field 11 = String "b95"

should work ?

Regards
Gag

[Martina]

Hm, yes, it does indeed, thanks for this idea (in fact one needs to indicate not the type of the variable, but String everywhere).

However I don't understand at all why I can write

Code: Select all

Scalar Field 10 = "b95"

but need to have

Code: Select all

Scalar Field 11 = String "b95"

?

And how did you see that it was linked to another Solver?

Martina

[gagliar]

Hi Martina,

Hm, yes, it does indeed, thanks for this idea (in fact one needs to indicate not the type of the variable, but String everywhere).

"String" is the type of the keyword (when reading in the sif, a string is expected).

And how did you see that it was linked to another Solver?

It was not linked to an other solver, but I had the same issue using an other solver

The explanation is in the file SOLVER.KEYWORDS in fem/src/ which contains all the listed keyword that can be found in the sif file, with their default type. If a keyword is in this list, there is no need to specify its type, else it is mandatory to specify the type (the case for your own solver for example).

Why it is working for 10 and not 11 is because the "nexp" is set to 10 in SOLVER.KEYWORDS. nexp is used for keywords which are defined on the form of a name + an integer, and these keywords are searched with a loop from 1 to nexp. It seems that nexp has been recently changed (was 32 in a previous version I have on my computer). Don't know why.

Anyway, I think it is a good practise to specify the type of all keywords...

See you
Gag

[Martina]

Hi Gag,

ok, thanks, now it's clear for me.
I agree, it's a better practise like that, I think it was just an old copy-pasteing from somewhere without really paying attention to that.

Martina